Computers are our friends?

December 10, 2014

Authorities from Arnold Schwarzenegger to Stephen Hawking have forewarned us of the dangers of intelligent machines. Yet we persist in further integrating computers in our life. Debating the role of computers in engineering is a pet topic of mine. Each time I return to the topic, I’ve softened my support for reliance on computational analysis.

Today I spent a couple of hours trouble-shooting a 3-D model for a roof truss system. Our problem stemmed from the computer’s assertion that second order P-delta effects could not be satisfied. In short, this is a phenomenon when a slight initial displacement itself results in greater moments, leading in turn to more displacement, and so on. Eventually, we traced the issue to some plausible causes, but in the process of “fixing” the model, I believe we introduced modeling constraints that could actually lead to a less conservative design. I’m not convinced the computer’s P-Delta algorithms were applied correctly to the complex system. After the exercise, I directed the engineer to create a simpler 2-D test model and do our own P-delta check based on the steel code’s hand check method. Alas, I may have to retire from the Emerging Engineer blog soon, as I now sometimes pine for the bygone days of 2-D modeling.

The lesson, as ever, is to use good engineering judgement and to question computer results. The balance of computer versus hand calculation remains pertinent today.

The below essay first appeared in Volume 10, Issue 2 (April 2010) of the Journal of Leadership and Management in Engineering. Unfortunately, the journal is no longer published. Fortunately, this means the entire catalog of back issues are free in the ASCE Library. All you have to do is register a user name and password; I don’t even think you need to be an ASCE member. I will occasionally republish my old columns, which appeared from 2010 through 2013.

In 2005, I wrote my very first paper for ASCE. I was intrigued by the ethical question posed for that year’s Daniel Mead Award for Younger Members: Is it ethical to use an engineering software program to solve a problem if you cannot complete the calculations manually? Embarking on a new assignment for ASCE, as a contributor to the Journal of Leadership and Management in Engineering, I felt compelled to return to that still-relevant topic. Five years of engineering practice have provided many more personal anecdotes that address the question and have forced me to consider the answer from a multigenerational perspective.

Still, my short and provocative answer to the question remains yes. I suspect many readers will chafe at that answer. Consider that the question can be read multiple ways. It may be a reliable Rorschach test for your relationship with computers and, by extension, a generational indicator. Did you interpret the question as, “Is it ethical to use software to solve a problem for which you don’t know how to do hand calculations?” or “…to solve a problem that cannot otherwise be answered by hand calculations?”

I have been involved in spreadsheet standardization efforts in several different companies. Each company’s relationship with spreadsheet calculations varied greatly. Some managers strongly resisted issuing an “approved” calculation sheet in the fear that novice engineers might use it in lieu of establishing a thorough understanding of the problem. Others feared institutionalizing a hidden error. Another type of manager saw the potential for increased efficiency and assurance of calculation quality. With such varied opinions, often within the same organization, spreadsheet standardization efforts rarely achieve their stated goals.

The managers on each side of the divide likely based their opinion of spreadsheets on personal experiences. Those with a cautious outlook probably had to deal with a situation in which an error was discovered at the eleventh hour. True believers have watched their groups’ productivity climb and have perhaps been lucky to avoid a major problem. Common ground lies in the interest in developing competent staff. Well-trained and conscientious engineers can be accurate and productive. In a final attempt to improve spreadsheet use, we placed standardization on the back burner and instead focused on developing the staff’s spreadsheet skills via presentations and the creation of a user group network.

Engineers face a different set of calculation problems when using advanced computer analysis methods on large, complicated systems. Even if one can claim to have a thorough understanding of matrix inversions and virtual work methods, some hand calculations remain outside the capacity of any professional working on a realistic schedule.

In one example, I was asked to peer review a reinforcement design proposed to address some cracked beams of a 50-story building. The problem appeared dire, because the spandrel beams in question carried the load of several dozen floors above. The consultant we reviewed had created a complex computer model of the exterior concrete moment frame using finite elements and nonlinear hinges to represent the cracked areas. We wanted to understand the redistribution of load to other beams in the system. However, both teams quickly learned that almost any favorable result could be obtained by modestly tuning the nonlinear hinges. Although many technical papers had been reviewed, the experienced engineers present had little confidence in any specific analytical recommendations.

Whereas young engineers might have been inclined to accept the computer analysis, the older generation understood that engineering judgment required a conservative approach. At the end of the project, we were still able to satisfy our client’s request to shave cost from the project. Experience was again trump, as many details of the design were recognized as redundant or inefficient. Taking some conservative data from the computer analysis, we even proposed alternative details to facilitate construction and minimize tenant disruption.

This is not to say that computer analysis always returns misleading, error-fraught results. In the analysis of a wind-damaged building, a team of interns quickly produced a complete finite element model of a 45-story steel moment-frame building. We wanted to know whether changes in the local wind climate, which might have caused the curtain-wall damage, would also influence the superstructure. The results indicated that some footings were close to overstress. Testing of the bearing conditions was conducted to verify safety. Building sway data, shared with wind tunnel consultants, helped also explain anecdotal stories of severe motion in high winds. Again, further on-site testing was performed to determine whether the issue was simply one of occupant comfort.

The interns building the model had only a basic understanding of finite element mechanics. However, they were trained on the interface of the modeling software, and their work was overseen by more experienced staff. Important assumptions, including those regarding applied loads and connection stiffness coefficients, were discussed with the most senior members of the team. The process went smoothly, even though the computer operators did not know how to do the calculations by hand and the senior staff members did not have the time to run the analysis by hand.

When I first answered this question five years ago, I was not yet able to apply these experiences to the formulation of my answer. I now believe that differences in (or lack of) experiences form the basis of the technology gap between generations. In these examples, I’ve shown the value of experiences gained by years of engineering practice. However, younger engineers also bring value from years of experience in computer operation. Other studies have quantified the value of multigenerational project teams, but it seems also to ring true for engineering analysis.

Younger engineers are no less concerned about their ethical responsibility to the public safety than older generations. The tools they use to perform their work may be different but still remain only a means to a calculation’s end. Older engineers ought not to dwell on the particulars of the tool. Regressing into a tale about a favorite slide rule is sure to cause those born into the computer age to tune out. Instead, share the story of training your peers, dropping the digital analysis to go with your engineering judgment, or integrating a multigenerational team. Good engineers of all ages want to do a good job efficiently. With your guidance, their experiences will eventually serve as their compass for ethical work.

Read More:

1 Comment
  • Good column, as usual Ken.

    Making others aware when the computer results are suspect or at least questioning the output is a noble effort in the first part and a necessary standard procedure on the second part. Making special note of the value different members on a team offer is even greater.

    Roll-back another generation+ and I consider the negative impact to me as an engineer that I never used a slide rule. I have had this friend named calculator. I never had to think through and figure out a decimal place for a result.


Leave a Reply

— required *

— required *